Terrain Classification

Traversal in unstructured outdoor environments requires knowledge of different terrain types so that the robot can decide between easily and not-so-easily traversible terrains. Visual classification in outdoor environments faces the difficulty of different lighting conditions. Different weather conditions and angle of the sun can make color information quite inaccurate for terrain recognition. We investigate techniques for robust and light invariant terrain classification.

We use our low-cost outdoor robot for this purpose. It is a custom built robot equipped with various sensors:

  • a single camera mounted on a pan tilt unit,
  • a laser range finder,
  • ultrasonic sensors, and
  • odometry.

Below, you can see a picture of the driving robot.

We also use the same approach on our flying robot. This is an AscTec Hummingbird quadrocopter fitted with a single camera flown at different heights on our campus.

The main research topics are:

  • Visual terrain classification
  • Terrain mapping.

Publications of our institute concerning this topic:

[1] Sebastian Otte, Stefan Laible, Richard Hanten, Marcus Liwicki, and Andreas Zell. Robust Visual Terrain Classification with Recurrent Neural Networks. In European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN), pages 451--456, Bruges, Belgium, April 2015.
[2] Andreas Masselli, Richard Hanten, and Andreas Zell. Localization of unmanned aerial vehicles using terrain classification from aerial images. In 2014 International Conference on Intelligent Autonomous Systems (IAS-13), Padova, Italy, July 2014. [ details ]
[3] Stefan Laible, Yasir Niaz Khan, and Andreas Zell. Terrain classification with conditional random fields on fused 3d lidar and camera data. In European Conference on Mobile Robots (ECMR 2013), pages 172--177, Barcelona, Catalonia, Spain, September 2013. IEEE. [ DOI | details | pdf ]
[4] Yasir Niaz Khan, Andreas Masselli, and Andreas Zell. Visual terrain classification by flying robots. In IEEE International Conference on Robotics and Automation (ICRA), pages 498 --503, St. Paul, Minnesota, USA, may 2012. [ DOI | details | pdf ]
[5] Stefan Laible, Yasir Niaz Khan, Karsten Bohlmann, and Andreas Zell. 3d lidar- and camera-based terrain classification under different lighting conditions. In Autonomous Mobile Systems 2012, Informatik aktuell, pages 21--29. Springer Berlin Heidelberg, 2012. [ DOI | details | link | pdf ]
[6] Yasir Niaz Khan, Philippe Komma, Karsten Bohlmann, and Andreas Zell. Grid-based visual terrain classification for outdoor robots using local features. In IEEE Symposium on Computational Intelligence in Vehicles and Transportation Systems (CIVTS 2011), pages 16 -- 22, Paris, France, apr 2011. [ DOI | pdf ]
[7] Yasir Niaz Khan, Philippe Komma, and Andreas Zell. High resolution visual terrain classification for outdoor robots. In Computer Vision Workshops (ICCV Workshops), 2011 IEEE International Conference on, pages 1014 --1021, Barcelona, Spain, nov 2011. [ DOI | pdf ]
[8] Philippe Komma and Andreas Zell. Posterior probability estimation techniques embedded in a bayes filter for vibration-based terrain classification. Springer Tracts in Advanced Robotics, 62:79--89, 2010.
[9] Philippe Komma and Andreas Zell. Posterior probability estimation techniques embedded in a Bayes filter for vibration-based terrain classification. In 7th International Conference on Field and Service Robots (FSR 2009), pages 1--10, MIT, Cambridge, Massachusetts, USA, July 2009. [ pdf ]
[10] Philippe Komma, Christian Weiss, and Andreas Zell. Adaptive Bayesian filtering for vibration-based terrain classification. In IEEE/RSJ International Conference on Robotics and Automation (ICRA 2009), pages 3307--3313, Kobe, Japan, May 2009. [ pdf ]
[11] Philippe Komma, Christian Weiss, and Andreas Zell. Improved vibration based terrain classification using temporal coherence. In 40th International Symposium on Robotics (ISR), pages 359--364, Barcelona, Spain, March 2009.
[12] Christian Weiss, Hashem Tamimi, and Andreas Zell. A combination of vision-and vibration-based terrain classification. In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2008), pages 2204--2209, Nice, France, 2008.
[13] Christian Weiss, Nikolas Fechner, Matthias Stark, and Andreas Zell. Comparison of different approaches to vibration-based terrain classification. In 3rd European Conference on Mobile Robots (ECMR 2007), pages 7--12, Freiburg, Germany, 2007. [ pdf ]
[14] Christian Weiss, Matthias Stark, and Andreas Zell. SVMs for vibration-based terrain classification. In Proc. Autonome Mobile Systeme (AMS), pages 1--7, Kaiserslautern, Germany, 2007. Springer. [ pdf ]
[15] Christian Weiss, Holger Fröhlich, and Andreas Zell. Vibration-based terrain classification using support vector machines. In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2006), pages 4429--4434, Beijing, China, 2006. [ pdf ]

Contact

Yasir Niaz Khan, Tel: (+49/0) 7071 / 29 78983, yasir.khan at uni-tuebingen.de
Stefan Laible, Tel: (+49/0) 7071 / 29 78983, stefan.laible at uni-tuebingen.de